A Survey of Latest Approaches for Crowd Simulation and Modeling Using Hybrid Techniques

Abstract

In recent past, computer simulation and modeling technologies have been used for variety of applications concerning crowds both in normal and emergency situations. Majority of models often fall into two extremes: either to simulate homogeneous high density crowd or to model individual social, physical and psychological behaviors of heterogeneous relatively small size crowd. Currently, researchers start exploring hybrid models that combine the benefits of both classical approaches to simulate large size heterogeneous crowds. In existing hybrid models, zones based in widely studied approach but have limited applicability due to its dependency on well defined boundaries of physical space. Other techniques like sequential and layer models need further investigations and provides limited crowd modeling features. Moreover, current hybrid models need to provide diverse behavioral features for crowd, further requires testing its impact on scalability and execution efficiency of simulation. There is significant need in literature for generic framework with flexible model selection capability to observe crowd dynamics. This paper assesses the major existing work in hybrid techniques for crowd modeling and simulation. We anticipate that this work will serve as useful insight on hybrid crowd modeling techniques, their issues and future research directions.